An Overview of Poultry Models - Nutritional Modelling for Pigs and Poultry - page 13

body weight gain at 77 days of age by +120.1,
-76.7, -94.5 and +60.9 g/day. At this age
daily gain is about 200 g/day so these are sig-
nificant levels of sensitivity. Given these
issues of estimation and sensitivity, it needs
to be clearly shown that model performance
is significantly improved by describing
growth at the level of protein turnover ra-
ther than considering just protein growth.
Finally the model of egg production and
nutrient response in broiler breeders de-
scribed by Gous and Nonis (2010) is included
in Table 1.2 . This is written at a similar level
of organization and uses similar principles to
the broiler models listed although, of course,
additional issues are raised in modelling egg
production. This model is discussed by Gous
(see Gous, Chapter 3, this volume).
The considerable level of accomplish-
ment and research revealed by these mech-
anistic approaches to modelling poultry
systems is self-evident. The application of two
of these models (see Méda et al ., Chapter 9,
this volume; Gous, Chapter 13, this volume)
was discussed at the 2013 International
Symposium of Modelling in Pig and Poultry
Production, but in general the uptake of
modelling in commercial decision making
in the poultry industry remains at a low level.
The commercial development of models is
useful in that it provides stimulus and fund-
ing but it is not helpful if the models are less
open, or in some cases, not open at all. There
is a commercial need to defend the model
and to prove it 'right', which contrasts with
the scientific need to criticize it and to
prove it 'wrong' so that it may be improved.
The scientific development of model-
ling concepts might be advanced more ef-
fectively by concentrating on the nature of
the problems that need to be resolved if suc-
cessful mechanistic models of poultry sys-
tems are to be developed. Such an approach,
rather than model development, might also
join together related work using empirical
trials, empirical modelling and the elabor-
ation of mechanistic ideas.
The poultry systems that we seek to
simulate vary quite widely but have a gen-
eral quality of being biologically complex.
The housing in one place of many, often
thousands, of animals creates a system of

great complexity that is difficult to describe.
Doing this in sufficient detail, so that differ-
ent commercial situations can be differenti-
ated but also in a way that can be reflected
in the principles of a model, is a first, and
very considerable, problem to be solved.
The model itself needs to be quite compre-
hensive if it is to be of use in practice. The
questions that need to be considered, for ex-
ample, for a broiler growth model, can be
grouped under three headings: the bird, the
environment and feed resources.
The description of the bird in terms of
its genetically determined potential per-
formance seems to be a powerful idea. For
broilers, and following Emmans (1981), in
EFG Software (1995) we use parameters for
potential feather-free body protein growth
and for a desired level of fatness. These
parameters can be determined in independ-
ent trials and, in principle at least, can be
adjusted to reflect genetic progress from
year to year. In addition, feathering must be
described and this presents a considerable
problem. Total feather growth (weight and
composition) can be incorporated in the
genotype description (e.g. see Gous et al .,
1999) but the description of feather cover,
which may be very uneven, is a problem
that still has to be solved. Similar ap-
proaches seem to be appropriate for turkeys
but the parameters for modern genotypes
remain to be determined. The description of
potential egg production is discussed by
Gous (see Gous, Chapter 3, this volume).
If models are to predict economic per-
formance they need to consider all those
components of the body that generate rev-
enue in different markets (Fisher and Gous,
2008). These will range from whole birds,
dressed carcasses and portions to dissected
meat. Offal and feet, for example may be
significant in some markets and 'waste'
components such as excess fat, may have
negative money value. The relationships
governing the growth of these many body
components will vary between breeds and
may be the subject of specific genetic ma-
nipulation as in the development of high-
yield broilers.
The description of the physical envir-
onment is superficially a simple problem,